The Agent Skills Directory

[SAFE]: The skill provides documentation and implementation patterns for Botpress ADK frontend integration. All identified external resources, including the @botpress/client package and the botpress.cloud domain, originate from the vendor's own infrastructure.
[INDIRECT_PROMPT_INJECTION]: The architecture described creates an attack surface where data from external sources is processed by a frontend application. 1. Ingestion points: Data enters the application through client.findTableRows() and client.callAction() calls as described in overview.md. 2. Boundary markers: None identified in the provided patterns. 3. Capability inventory: The application performs UI rendering and state management. 4. Sanitization: The documentation recommends using the sanitize-html library and provides best practices for secure rendering in recommended-stack.md.
[SAFE]: The skill illustrates patterns for handling authentication via Personal Access Tokens (PATs). It recommends storing these tokens in cookies with the SameSite=Lax attribute for protection and includes instructions to avoid hardcoding secrets or logging tokens in client-side code, as outlined in authentication.md.

adk-frontend